Improving the uncertainty assessment of economic losses from large destructive wildfires

Abstract

Currently, as fire risk is considered a high-frequency and low-severity risk, actuarial and underwriting pricing and risk management methods have stuck to methods based purely on historical loss data. In the global context of both increasing fire severity with climate change and increasing wildland–urban interface area, the use of environmental-based dynamical modelling tools offers a good alternative to better evaluate fire risk. A new method is presented here that combines the raster-based fire spread model SWIFFT and a stochastic approach for generating the spatial and temporal distribution of ignition points. Monte Carlo simulations are conducted, and the uncertainties of hazard and losses are evaluated. Adapted and applied to the landscape conditions of a Brazilian plantation, it is shown to be well suited for a precise reconstruction of the fire burnt area. Finally, the uncertainty assessment of losses for this study zone is presented. We conclude by discussing this new method, which has a high level of traceable uncertainty and how fire risk insurance can deal with it, as well as the progress of future research that will benefit from this method.